Evaluation of password hashing schemes in open source web platforms

Abstract

Nowadays, the majority of web platforms in the Internet originate either from CMS to easily deploy websites or by web applications frameworks that allow developers to design and implement web applications. Considering the fact that CMS are intended to be plug and play solutions and their main aim is to allow even non-developers to deploy websites, we argue that the default hashing schemes rarely are modified. Also, recent studies suggest that even developers do not use appropriate hash functions to protect passwords, since they may not have adequate security expertise. Therefore, the default settings of CMS and web applications frameworks play an important role in the security of password storage. This paper evaluates the default hashing schemes of popular CMS and web application frameworks. First, we formulate the cost time of password guessing attacks and next we investigate the default hashing schemes of popular CMS and web applications frameworks. We then apply our framework to perform a comparative analysis of the cost time of password guessing attacks between the various CMS and web application frameworks. Finally, considering that intensive hash functions consume computational resources, we analyze hashing schemes from a different perspective. That is, we investigate if it is feasible and under what conditions to perform slow rate denial of service attacks from concurrent login attempts. Through our study we have derived a set of critical observations. We have discovered that many CMS and web application frameworks use outdated hash functions, arbitrary number of hash iterations, while there is a lack of password policies and salt. Notably, the popular WordPress still uses MD5 with low number of hash iterations. Overall, we believe that the security status of the hashing schemes of CMS and web application frameworks calls for changes to the default settings from an opt-in to an opt-out security policy. More security audits and official library implementations are also required to accelerate the adoption of memory hard functions both by policy makers and the industry.